Rubymine and Junie : Dejavu all over again.

A major reason for why we only set a general indicator for your quota is JetBrains AI provides multiple AI models with different capabilities and prices.

For example, o3 series models will use more quotas than gpt-4o-mini. At the current stage, the quota consumption is dynamically adjusted based on the entire usage. We plan to add the detailed quota consumption in the next update, when we have more data of the entire resource usage.

You could consider upgrade to AI Ultimate subscription to get more quota of cloud LLM resources. In case you reach the quota limitation after upgrading to AI Ultimate subscription, please contact us, and we will help to resolve the issue.

I have upgraded to Ultimate, reluctantly, to get a particular job finished. 

The reason I mentioned “deja vu” is this is a repeat of the initial AI launch where your marketing estimate and the real life experience massively overestimated what the plan allowed and people ran out within days and had no options till renewal or go through a reset process.

I am currently using Claude for most coding that used to time me out every few hours ( not till the end of the month ) which was fine. Now I despite using it even more than before it no longer times me out. Very different experience.

Still like your stuff, just not the over-estimation of what each AI plan allows. In reality I used it for 3 days because I took a time out due to the warning. Not a great user experience.

I understand your situation. When AI Pro was released, AI Assistant didn't have any agentic coding features. We added Edit mode and released Junie in April, it works well, but the previous LLM credits cannot meet the agentic coding requirements. We have to provide new pricing plans for JetBrains AI (in most of the use cases, the cost of LLM services still far exceeds AI Pro/Ultimate price). 

If you have any questions on using Junie or feedback on JetBrains AI, please feel free to reach out us.